Do Search Engines and AI Understand Your Business?
You know what your business does.
You know what you’re particularly good at, the problems you solve, the questions customers ask, the things that can go wrong and the lessons you’ve learned from years of actually doing the work.
Your team knows even more.
But how much of that knowledge exists outside your collective heads?
If we removed everything you already know about your own business and looked only at the information available online, would we come to the same conclusions about what your business does, what you know and what you’re genuinely good at?
Because search engines and AI systems can’t pop into the office for a coffee and ask.
They have to work with the information available to them.
And there can be a surprisingly large gap between what a business genuinely knows and does, and what its digital presence demonstrates that it knows and does.
That gap has always mattered to the people looking for your products or services.
Now that search engines and AI systems are also trying to interpret businesses, subjects, expertise and the relationships between them, it matters in another way too.
Imagine You Own a Bakery
Let’s make this ridiculously simple.
You own a bakery and you’re brilliant at wedding cakes.
You’ve made hundreds of them.
You know which fillings work well in tiered cakes, how to support the structure, how many portions different sizes provide and which designs survive being transported.
You know what happens to buttercream in hot weather. You understand dietary requirements. You know how far ahead couples should order and which seemingly wonderful ideas are likely to end in disaster.
There’s a decade of knowledge sitting inside your head.
Your website says:
We make cakes for all occasions.
That’s it.
A potential customer looking at your website doesn’t get to see most of your expertise.
Neither does a search engine.
Neither does an AI system trying to establish which businesses might be useful when somebody asks about ordering a wedding cake.
You have the knowledge. Your website doesn’t.
And search engines and AI can’t read your mind.
Yet. (Only joking. Or am I?)
Your Website Isn’t Your Business
That sounds obvious, but it’s an important distinction.
A website is a representation of a business.
It is not the business itself.
We recently encountered a fairly spectacular example of this with one of our clients.
We hadn’t been working with the company for particularly long when it became obvious that some of its services were missing from the website.
Nothing particularly unusual there. Businesses evolve. Services get added. Websites aren’t always updated as quickly as the businesses behind them.
So we started adding the missing service pages.
Then we encountered another problem.
As more services were added, the website menu started becoming unwieldy. It needed restructuring.
We suggested creating top-level service categories and asked the team to do something that sounded very simple:
List all the services the company provides, and we’ll work out how they should be grouped.
Sometimes I rue the questions I ask.
The list came back.
There were 47 services completely missing from the website.
Not obscure things the company could theoretically provide.
Services it provides all the time.
Once those 47 services were grouped into 10 logical top-level categories, we were looking at 57 new pages required to accurately represent what this business actually does.
Fifty-seven.
Now, this is an extreme example. Most businesses aren’t going to discover nearly five dozen missing pages when somebody finally asks the right question.
But it rather neatly proves the point.
The company hadn’t suddenly acquired 47 new services.
The expertise was already there. The work was already being done. The website simply didn’t know about it.
And if the company’s own website didn’t contain that information, we’d made it unnecessarily difficult for potential customers, search engines and AI systems to discover the full extent of what the business actually did.
That’s the gap we’re talking about.
And this isn’t merely a technical search problem.
If somebody lands on your website looking for a service you provide every week, but that service isn’t there, they don’t know that you forgot to add the page.
They assume you don’t do it.
Then they go somewhere else.
Search Engines and AI Need Evidence to Understand Your Business
Your website isn’t the only place information about your business exists.
Search engines and AI systems may encounter information about a company through reviews, directories, social platforms, news coverage, industry websites, customer websites and other published sources.
Different search engines and AI systems also work in different ways. Some information may come from a search index, some from external retrieval and some from knowledge acquired during an AI model’s training.
So when we talk about what search engines and AI systems “understand”, we’re not suggesting there’s one enormous machine sitting somewhere on the internet building a definitive profile of your company.
Nor are we suggesting that an AI system sits contemplating your business over a cup of tea.
We’re using understand in a much more practical sense:
Is there enough clear, accessible information for these systems to identify what your business is, what it does, what it knows and when it might be relevant?
Your own website matters enormously within that picture because it’s the place where you control the explanation.
A company might have ten people with a combined 150 years of experience.
Those people might solve complicated problems every day. They answer highly specific customer questions, spot problems inexperienced people miss and make decisions based on years of actually doing the job.
Then you visit the website.
There are a handful of service pages, an About page written six years ago and some fairly generic articles that could have been written for almost any company in the industry.
The knowledge inside the business and the knowledge demonstrated online are two very different things.
This is where the conversation becomes bigger than simply making sure every service has a page.
Because being mentioned isn’t the same as being understood.
Suppose your website does have a page about a particular service.
Does that page demonstrate genuine knowledge?
Does it explain the problems customers encounter?
Does it answer their questions?
Does it show experience?
Does it connect the service to relevant problems, products, sectors or other services?
Does it provide evidence?
Or does it essentially say:
We provide this service. We’re very experienced. Contact us for a quote.
That’s better than not mentioning the service at all.
But it still leaves a lot of knowledge trapped inside the business.
Interestingly, this isn’t something we have invented simply because AI search is fashionable. Google’s own guidance asks whether content demonstrates first-hand expertise and depth of knowledge, whether it contains original information or analysis and whether someone reading it would leave feeling they had learned enough to achieve their goal.
Google’s newer guidance specifically covering generative AI search continues that theme, encouraging businesses to create useful, non-commodity content based on what they genuinely know rather than simply recycling information that already exists online.
So the direction isn’t particularly mysterious.
Know something useful. Explain it properly. Demonstrate why you’re qualified to explain it.
How Does a Machine Work Out What Your Business Is About?
You don’t need to understand the technology behind this to benefit from it, but understanding one simple idea helps explain why these gaps matter.
This is where marketing terminology can make a relatively straightforward idea sound much more complicated than it needs to be.
One of the words you’ll encounter is entity.
An entity is essentially an identifiable thing or concept.
That might be a business, person, place, product, service or subject.
Imagine a commercial heating and cooling company.
A machine analysing information about it might encounter things such as:
Commercial heating
Air conditioning
Ventilation
Maintenance
Energy efficiency
Commercial buildings
Facilities managers
Milton Keynes
Identifying those individual things is useful.
But understanding the relationships between them is even more useful.
The company provides commercial heating maintenance.
Air conditioning and ventilation relate to building climate control.
The company works with commercial buildings.
The company operates in Milton Keynes.
This is subtly different from the old-fashioned idea of looking at a page and counting how many times a particular keyword appears.
We’re trying to understand what the information means and how the pieces fit together.
Where Does Schema Fit Into This?
Schema, or structured data, is another term that frequently appears in conversations about search and AI.
Again, the basic idea is much simpler than the terminology suggests.
Imagine you’ve drawn a picture of a dog.
A person looks at it and thinks:
That’s a dog.
A machine can analyse the information and try to reach the same conclusion.
Schema is a little like sticking a clear label underneath saying:
DOG.
It provides structured clues about what something is and, in some cases, how it relates to other things.
Google describes structured data in much the same way, as a standardised format that provides explicit clues about the meaning of a page and classifies its content.
That’s useful.
But sticking a label saying DOG underneath a picture of a hamster doesn’t magically turn it into a dog.
Schema doesn’t manufacture expertise either.
It can help machines interpret information that exists. It cannot compensate for information that doesn’t.
And this is where we need to separate what Google actually says from some of the more enthusiastic claims currently being made about AI optimisation.
Google recommends structured data where appropriate and uses it to help understand page content and support certain search features.
But its current guidance for generative AI search doesn’t introduce a magical new piece of “AI schema” that businesses need to add in order to appear.
So yes, structured data can be useful.
It isn’t a substitute for actually explaining what your business knows and does.
The Idea That Started This Article Was Actually Much Narrower
This whole train of thought began when we encountered an interesting approach to content gap analysis.
The suggestion was essentially this:
Analyse the entities associated with a website.
Compare them with the entities associated with the wider subject.
Look at competitors.
Find the gaps.
Then use those gaps to identify content opportunities.
Clever.
And there is something genuinely useful in it.
But there was something bothering us.
What if the biggest gap isn’t between your website and your competitors?
What if it’s between your website and your own business?
That’s when the idea became much more interesting.
Because before we start analysing what everybody else knows, perhaps we should establish whether our own digital presence accurately represents what we know.
The more we explored it, the more obvious it became that the gap is much bigger than entities.
The Interesting Question Is What’s Missing
Go back to our imaginary commercial heating and cooling company.
Its website talks extensively about air conditioning installation and commercial boilers.
When we examine the wider subject, we encounter things such as planned preventative maintenance, ventilation, energy efficiency, indoor air quality and system controls.
Some are barely mentioned on the website.
Have we just discovered five new articles to write?
Absolutely not.
We’ve discovered five questions to ask.
Does the company actually provide these services?
Does it genuinely have expertise in these areas?
Do its customers care about them?
Do they relate to commercially important work?
Are people looking for information about them?
Are they already covered elsewhere using different terminology?
Can the company contribute something genuinely useful?
If the answer is no, the gap might not matter.
But imagine the business says:
“Planned preventative maintenance? Yes. That’s one of our most important services. We’ve been doing it for twenty years.”
Now we have something interesting.
Why couldn’t we tell that from the website?
That’s a meaningful gap.
Try This With Your Own Website
You don’t need sophisticated software to start testing this.
Write down the five things you most want your business to be known for.
Not the five things you think people search for.
The five things that matter commercially to your business.
Now go to your website and pretend you know nothing about the company.
Can you find each one easily?
Is there a page explaining it properly?
Would somebody understand why you’re good at it?
Can you find evidence that you’ve actually done it?
Does the website answer the questions a potential customer would probably have?
Now remove everything you personally know about the business and rely entirely on what the website tells you.
Would you still conclude that those five things were important areas of expertise?
If not, you’ve just found a gap.
We Could Make This Enormously Complicated
We could crawl thousands of URLs.
We could extract entities, build knowledge graphs, analyse semantic relationships and produce some very impressive spreadsheets.
Eventually, we probably will.
But before we do any of that, there are five fairly ordinary questions worth answering.
1. The Business Map: What Do You Actually Know and Do?
Start with the business.
Not Google.
What services do you provide?
What products do you sell?
What problems do you solve?
Which customers and sectors do you understand particularly well?
What are your genuine areas of expertise?
What do the people inside the company know that isn’t written down anywhere?
What has experience taught you?
What do customers regularly ask your team?
What decisions do your experienced people make almost instinctively because they’ve encountered the problem a hundred times before?
And importantly:
What does the business actually want to be known for?
Not rank for.
Be known for.
This exercise alone can be surprisingly revealing.
Remember those 47 missing services?
They weren’t discovered by an SEO tool.
They appeared because somebody finally asked the business to list what it actually does.
2. The Website Map: What Does Your Website Demonstrate?
Now forget everything you know about the company and examine the website.
What would somebody encountering this business for the first time conclude?
Which services are clearly explained?
Which areas of expertise receive substantial coverage?
Which are barely mentioned?
Which don’t appear at all?
Does the website answer the questions customers actually ask?
Does it explain relevant problems and solutions?
Are related subjects connected?
Does it demonstrate experience, or merely claim it?
The difference between the Business Map and the Website Map can expose some enormous holes before we’ve looked at a single competitor or keyword.
And sometimes those holes will be embarrassingly obvious.
We sell this. The website doesn’t say we sell this.
Others will be subtler.
We say we do this, but nothing on the website demonstrates that we understand it particularly well.
Both matter.
3. The Subject and Search Map: What Else Belongs in the Conversation?
Now we look outwards.
If somebody genuinely wanted to understand the subject, what would they need to know?
What questions do customers ask?
What problems do they encounter?
What decisions do they need to make?
What concepts are naturally related?
What terminology is used?
What are people actually searching for?
This isn’t about discovering another hundred keywords to shoehorn into the website.
It’s about understanding the whole subject, rather than a collection of search phrases.
And this is where entity and relationship analysis becomes genuinely useful.
It can help us identify parts of a subject we may not have considered.
But identifying them isn’t the same as deciding they’re relevant.
That’s still our job.
4. The Competitive Map: What Might Others Have Spotted That You Haven’t?
Competitor analysis can be useful.
If several credible competitors have all covered an important aspect of a service that your website ignores, that’s worth investigating.
It doesn’t automatically mean you should cover it too.
Perhaps it isn’t relevant to your business.
Perhaps you deliberately don’t provide it.
Perhaps your competitors are all copying each other and nobody needed the content in the first place.
But perhaps they’ve identified an important customer question, service or aspect of the subject that you’ve overlooked.
The competitor has given us a question, not an instruction.
That’s an important distinction.
We aren’t trying to create a slightly larger version of a competitor’s website.
We’re using their digital presence as another source of evidence about what might be missing from ours.
5. The Evidence Map: Can You Prove It?
Finally, look at the evidence.
It’s very easy for a website to say:
We’re experts in X.
What supports that claim?
Do you have case studies?
Real examples?
Customer reviews?
Photographs?
Qualifications?
Named experts?
Original insight?
Detailed explanations based on actually doing the work?
Processes developed through experience?
This matters to people first.
If I’m considering giving a company a significant piece of work, “we’re experts” isn’t particularly persuasive on its own.
Show me.
It also happens to align with the way Google talks about useful content. Its guidance specifically encourages clear sourcing, evidence of expertise, information about who created content and original information or analysis.
There is a significant difference between claiming expertise and demonstrating expertise.
The latter gives customers, search engines and AI systems considerably more evidence to work with.
Not All Gaps Are the Same
Once we put those maps together, different types of gap begin to appear.
A communication gap exists when the business knows or does something important but its website barely mentions it.
A depth gap exists when a subject is mentioned but the content doesn’t demonstrate meaningful knowledge.
An evidence gap exists when expertise is claimed but there is little to substantiate it.
A relationship gap occurs when useful pieces of information exist but the connections between them aren’t clear.
A search gap exists when customers have relevant questions or needs that the website isn’t addressing.
A competitive gap can highlight relevant territory credible competitors cover that the business has overlooked.
A structural gap occurs when useful information exists but is buried, isolated or poorly connected to the rest of the website.
And there can be an understanding gap, where the information technically exists but isn’t communicated clearly enough for either people or machines to easily make sense of it.
Different gaps require different solutions.
You don’t need to remember the names.
The important point is that “we need more content” is far too simplistic a diagnosis. First we need to understand what’s actually missing, why it matters and what would genuinely improve the information available to the customer.
Which brings us to an important point.
Finding a Gap Does Not Mean Writing Another Bloody Blog
This is where content gap analysis can go horribly wrong.
An SEO tool says a competitor ranks for something you don’t.
Another tool says people search for it.
Therefore:
ARTICLE!
Not necessarily.
Before creating anything, ask whether the gap actually matters.
Does it relate to genuine expertise?
Is it relevant to customers?
Does it support something commercially valuable?
Is there a real information need?
Can the business contribute something useful?
If the answers are no, we can do something marketing departments occasionally struggle with.
Nothing.
Not every gap needs filling.
Not every keyword deserves a page.
Not every related entity needs shoehorning into an article.
Not every customer question requires 1,500 words.
And filling a website with content simply because a tool identified an opportunity is a very good way to create a large website without creating a useful one.
Google’s own guidance is worth noting here. It warns against producing lots of content across many topics simply in the hope that some of it performs in search, and its newer generative AI guidance stresses useful, original, non-commodity content rather than mass-produced variations of information that already exists.
That’s Google’s position.
Our marketing conclusion is simpler:
Just because you can create a page doesn’t mean you should.
Sometimes the Answer Isn’t New Content at All
Even when we’ve identified a meaningful gap, the answer isn’t necessarily a new page.
Perhaps an existing service page needs expanding.
Maybe two existing pages need connecting properly.
Perhaps the missing piece is an FAQ based on something customers regularly ask.
Maybe there’s an excellent case study sitting in somebody’s inbox that demonstrates exactly the expertise we’re trying to establish.
Perhaps the website needs better information about the people delivering the service.
Maybe existing content needs clearer internal linking.
Structured data might help clarify information that’s already there.
Or perhaps there genuinely is a subject substantial and useful enough to warrant a new article.
The gap identifies the problem. Strategy decides what to do about it.
Technology Can Help Us Find What We’ve Missed
This is where some of the clever technology becomes genuinely useful.
We can crawl websites.
We can analyse recurring subjects and concepts.
Natural language processing can help identify entities.
We can analyse structured data.
We can examine search behaviour.
We can compare websites with competitors.
We can explore relationships between subjects.
And AI can help process quantities of information that would take a human an extraordinary amount of time to work through manually.
That’s incredibly useful.
But we shouldn’t confuse the tool with the strategy.
A machine might identify:
This subject appears frequently over there and rarely over here.
That’s information.
Deciding whether that subject reflects genuine expertise, matters to customers, supports commercial objectives and deserves greater prominence is a different job.
That’s judgement.
And this is why the original entity-gap idea only takes us part of the way.
The technology can help us spot the hole.
The business context tells us whether it’s a hole worth filling.
AI Changes the Way People Can Ask the Question
This is where the relationship between all these concepts becomes particularly interesting.
Traditional search has largely trained businesses to think in terms of searches that look something like:
commercial air conditioning maintenance Milton Keynes
There’s a service.
There’s a location.
We can understand why somebody might search for it and create useful information accordingly.
But someone using a generative AI system might ask:
“We manage three office buildings around Milton Keynes and the air conditioning keeps failing during hot weather. We normally call somebody when it breaks. Should we be looking at planned maintenance instead?”
Look at how much more is going on inside that question.
Commercial buildings.
Air conditioning.
Recurring faults.
Hot weather.
Reactive repairs.
Planned preventative maintenance.
Facilities management.
Location.
The person isn’t necessarily asking for a page containing one exact phrase.
They’re describing a situation.
For a system to produce a useful answer, it needs to interpret the relationships between those different parts of the problem and retrieve or use information relevant to them.
And if it moves from answering the question to helping that person investigate possible suppliers, the businesses with clear, accessible evidence connecting those areas give the system more useful information to work with.
That does not mean:
Mention all the entities and ChatGPT will recommend your company.
If anybody promises you that, keep hold of your wallet.
It means something much less exciting and much more defensible:
Clearly explaining what you do, what you know, the problems you solve and how those things relate gives both people and machines better information with which to assess your relevance.
What Does Any of This Have to Do With AI Search?
Quite a lot.
But probably not for the reason you might expect.
There is currently no shortage of advice promising special tricks for “optimising for AI”.
Some of it is useful.
Some of it is old SEO wearing a new hat.
And some of it is considerably more confident than the available evidence justifies.
Google’s current guidance for AI search makes an important point: existing SEO best practices remain relevant to its generative AI features, with an emphasis on valuable, original, non-commodity content.
That’s worth paying attention to.
But we should also be careful not to turn Google’s explanation of Google’s AI search products into a universal explanation of every AI system.
They don’t all work identically.
That’s another reason we’re much more interested in the underlying marketing principle than in chasing a particular technical trick.
People now use traditional search engines and generative AI systems to find information, investigate problems, compare options and research businesses.
Those systems need evidence from which relevance can be established.
Making your genuine expertise clearer, deeper, better connected and better evidenced gives people and machines better information with which to interpret your business.
That’s not an AI hack.
It’s good digital marketing.
And importantly, it doesn’t require us to abandon everything we already know about creating useful websites and content.
Quite the opposite.
The Last Word
There are some extraordinarily sophisticated tools available to help us understand what search engines and AI systems might make of a website.
We can extract entities, analyse semantic relationships, examine structured data, interrogate search behaviour, crawl competitors and use AI to process thousands of pages.
And all of that can be useful.
But before we do any of it, there’s a wonderfully unsophisticated question worth asking:
Does your website actually show what your business knows and does?
One of our clients was missing 47 services from its website.
No algorithm discovered them.
We asked the business what it did.
That’s perhaps the most important lesson in all of this.
Search and AI optimisation shouldn’t begin with trying to work out what machines want to hear.
It should begin with understanding what is true about the business, what matters to its customers and whether its digital presence accurately demonstrates it.
Then technology can help us find the subtler gaps.
The expertise that’s mentioned but never properly explained.
The claims that aren’t supported by evidence.
The questions customers ask that nobody has answered.
The relationships between services and subjects that haven’t been made clear.
The useful information that’s buried somewhere on the website but disconnected from everything around it.
The valuable knowledge that’s still sitting inside somebody’s head instead of helping the person searching for it.
And that’s why we think the idea of an “entity gap” is useful, but too small.
The real gap is between what your business genuinely knows and does and the evidence of that knowledge available to the outside world.
Sometimes that gap will require sophisticated analysis to find.
Sometimes it will require schema.
Sometimes it will require search data, competitor research or AI.
And sometimes you’ll find 47 missing services because somebody finally thought to ask the bloody question.
The objective isn’t to convince a search engine or an AI system that your business knows more than it does.
It’s to stop your website making it look like your business knows less than it does.
Part of the Marketing Clarity Series
This article is part of the Marketing Clarity series from The Last Hurdle, exploring the thinking behind clearer, more effective marketing.
From visibility and customer journeys to AI discovery and meaningful measurement, the series looks beyond the numbers to consider whether your marketing accurately represents your business, reaches the right people for the right reasons, and contributes to what you’re actually trying to achieve.
